Peripheral Plasmodium falciparum Infection in Early Pregnancy Is Associated With Increased Maternal Microchimerism in the Offspring

Abstract

Background

Placental malaria has been associated with increased cord blood maternal microchimerism (MMc), which in turn may affect susceptibility to malaria in the offspring. We sought to determine the impact of maternal peripheral Plasmodium falciparum parasitemia during pregnancy on MMc and to determine whether maternal cells expand during primary parasitemia in the offspring.

Methods

We conducted a nested cohort study of maternal-infant pairs from a prior pregnancy malaria chemoprevention study. Maternal microchimerism was measured by quantitative polymerase chain reaction targeting a maternal-specific marker in genomic DNA from cord blood, first P falciparum parasitemia, and preparasitemia. Logistic and negative binomial regression were used to assess the impact of maternal peripheral parasitemia, symptomatic malaria, and placental malaria on cord blood MMc. Generalized estimating equations were used to assess predictors of MMc during infancy.

Results

Early maternal parasitemia was associated with increased detection of cord blood MMc (adjusted odds ratio = 3.91, P = .03), whereas late parasitemia, symptomatic malaria, and placental malaria were not. The first parasitemia episode in the infant was not associated with increased MMc relative to preparasitemia.

Conclusions

Maternal parasitemia early in pregnancy may increase the amount of MMc acquired by the fetus. Future work should investigate the impact of this MMc on immune responses in the offspring.

The impact of maternal Plasmodium falciparum infection on cord blood microchimerism was investigated using dried blood spots from a Ugandan cohort. Early but not late maternal parasitemia was associated with increased detection of maternal microchimerism.

During pregnancy, rare maternal cells traffic from the mother into the fetus, a phenomenon known as maternal microchimerism (MMc) [1]. These cells have been identified in the fetus as early as 13 weeks gestation [2], where they are resident in the fetal lymphoid tissue [3] and other organs and can be easily identified in cord blood at the time of delivery [4]. Maternal microchimerism is present in immune and nonimmune cells throughout the body [1], and in cord blood the highest levels have been found in antigen-experienced T cells [5]. In addition, the acquisition of MMc is associated with the development of fetal regulatory T cells that induce tolerance against noninherited maternal alloantigens [3]. The potential beneficial or detrimental effect of MMc in the offspring is little studied. In humans, limited data suggest that “mother of the cord” cells may have an antitumor effect in cord blood transplant recipients [6], and one report documents the expansion of maternal cytomegalovirus-specific T cells in an immunocompromised infant [7]. In contrast, other work has found an association between increased MMc and pre-term delivery [8].

Although the determinants of the size and cellular phenotype of the maternal graft are largely unknown, we have previously shown in a cohort from Tanzania that inflammatory placental malaria was associated with higher levels of MMc in the cord blood of the offspring [9]. Recent literature suggests that Plasmodium falciparum infection early in pregnancy may impact placental development even in the absence of placental malaria at delivery [10], but the impact of peripheral P falciparum parasitemia during pregnancy on the transfer of maternal cells to the fetus is unknown.

Further, in our prior work cord blood MMc was associated with increased risk of P falciparum parasitemia but decreased risk of symptoms in infected infants [9]. Although the mechanisms driving this modulated malaria susceptibility in infants remain unclear, one possibility is that in malaria-endemic settings infants may acquire maternal, malaria antigen-specific T cells. Alternatively, regulatory T cells responding to persistent maternal alloantigen may induce cross-tolerance to malaria antigen in the offspring [11–13].

In this study, we examined the impact of early (12–20 weeks gestation) and late (>20 weeks gestation) P falciparum peripheral parasitemia on the presence of MMc in cord blood. We hypothesized that early peripheral infection occurring during initial placental implantation and spiral artery remodeling might have a disproportionate effect on MMc acquired by the fetus. We also compared the prevalence and level of MMc in the offspring in paired samples obtained just before and at the time of first P falciparum infection in infants. We hypothesized that if the offspring of malaria-exposed infants acquire a graft containing malaria-specific maternal T cells, these cells may expand upon encountering their cognate malaria antigen during primary malaria infection in the infant.

METHODS

Cohort and Plasmodium falciparum Exposure

Maternal-infant pairs considered for inclusion in the nested study were derived from PROMOTE, a previously conducted trial of chemoprophylactic regimens for intermittent preventive therapy during pregnancy to reduce the risk of placental malaria conducted in Tororo, Uganda [14]. Pregnant women were enrolled between 12 and 20 weeks’ gestation and observed every 4 weeks throughout the pregnancy. The present study evaluated women enrolled in the sulfadoxine-pyrimethamine chemoprophylaxis arm of the trial who received doses at 20, 28, and 36 weeks’ gestation and who delivered at or after 35 weeks’ gestation. Plasmodium falciparum parasitemia during febrile visits (symptomatic infection) was assessed from thick blood smears by microscopy. Parasitemia (asymptomatic infection) during routine visits was assessed from dried blood spots (collected at enrollment and 24, 28, 32, 36, and 40 weeks’ gestation) using loop-mediated isothermal amplification (LAMP). Symptomatic malaria infections were treated with artemether-lumefantrine. Placental malaria was defined as the presence of any parasites or malaria pigment detected by histopathologic evaluation of the placenta. All women were negative for human immunodeficiency virus and filaria. Offspring were subsequently observed for the first 3 years of life with routine visits every 4 weeks and were randomized to receive monthly versus every 3 months dihydroartemisinin-piperaquine between 2 months and 2 years of age. Caregivers were encouraged to bring children to the study clinic whenever they were ill. Institutional review board approval was provided by the Uganda National Council of Science and Technology, the University of California, San Francisco, Makerere University, and Seattle Children’s Hospital.

Deoxyribonucleic Acid Extraction and Maternal Microchimerism Measurement

Genomic deoxyribonucleic acid (DNA) was extracted from dried blood spots from maternal peripheral blood at delivery, cord blood, and infant blood samples using an optimized Chelex resin-based approach [15]. Mothers and infants were human leukocyte antigen class II genotyped by direct sequencing (Scisco Genetics, Seattle, WA) to identify a unique maternal marker for each pair. To quantify the microchimeric genomic equivalents (gEq) present in cord blood and infant samples, the maternal specific marker for each pair was amplified using a panel of previously validated quantitative polymerase chain reaction (qPCR) assays [5, 16] with slight modification. To ameliorate the effect of potential PCR inhibitors found in dried blood spots, PCR reaction mixes were additionally supplemented with bovine serum albumin (Thermo Fisher Scientific, Waltham, MA). Each assay is sensitive to detect a single amplicon in the background of 60 000 gEq [17]. A calibration curve for the maternal-specific assay was included to quantify the amount of MMc for each experiment. Every sample was also tested for the nonpolymorphic β-globin gene, and a β-globin calibration curve (human genomic DNA [Promega]) was concurrently evaluated on each plate to quantify the total number of gEq of DNA tested from each sample. Each sample was classified as positive or negative for MMc based on the presence or absence of any amplification of the maternal-specific marker. The level of MMc was determined by comparison of the amplification of the maternal marker in the sample against the assay-specific calibration curve, and it is expressed as the ratio of microchimeric gEq per 100 000 total gEq tested.

Statistical Analysis

Detection of MMc at birth was modeled using logistic regression, adjusting for gEq tested. The level of MMc at birth was modeled using negative binomial regression, including parameters for the microchimeric gEq count and β-globin gEq count detected in each sample. Negative binomial modeling was used becauses it accounts for the large number of zeros in the dataset and the variable gEq assessed in each subject [5, 9, 17, 18]. The output of the model is a ratio of normalized microchimeric gEq in the comparator group versus the reference group, which is presented here as the detection rate ratio (DRR). For example, a DRR value of 5 is interpreted as 5 microchimeric gEq detected in the comparator group for every 1 microchimeric gEq in the reference group.

To assess the effect of maternal peripheral parasitemia on cord blood MMc in the infant, LAMP positivity at enrollment (12–20 weeks gestation, “early parasitemia”), LAMP positivity after 20 weeks gestation (“late parasitemia”), symptomatic malaria, and placental malaria were first included in univariate analyses. Associations between early and late parasitemia, symptomatic malaria, and placental malaria were assessed using the χ ² test. Subsequently, because early and late malaria were not associated, they were included together in the adjusted models to estimate an independent effect of each adjusted for the other. Placental malaria was not included in adjusted models because it was associated with both early and late parasitemia. Covariates considered for inclusion in the adjusted models included gravidity, estimated gestational age, infant sex, and history of indoor residual spraying (IRS) before delivery. Covariates were included in the final adjusted models if they functioned as a predictor (P ≤ .1) or confounder of the relationship between early parasitemia and either detection or level of MMc (changing the estimate by 10% or more when added to the model). Gravidity (predictor, confounder) and IRS (confounder) were included in the adjusted analyses of cord blood MMc. Comparison of detection and level of MMc between cord and preparasitemia samples to assess change over time and between preparasitemia and parasitemia samples was conducted using generalized estimated equations with either a binomial or negative binomial structure. Adjusted analyses included early and late maternal parasitemia and infant parasitemia (primary variables of interest) and IRS (confounder). Estimation of the relationship between gravidity and infancy MMc could not be conducted using logistic regression because there was no MMc amongst primigravidae; thus, distribution was assessed with the χ ² test. All statistics were conducted in Stata 14.

RESULTS

Demographics and Malaria Exposure

Maternal-infant samples from delivery were available for 81 pairs among whom there was an informative marker to measure MMc in the offspring for 57 pairs (70%). Women spanned a broad age range, approximately one third were primigravidae, and almost two thirds reported prenatal IRS exposure (Table 1). Eighteen (32%) women experienced symptomatic malaria and 31 (54%) had placental malaria. Plasmodium falciparum parasitemia during early pregnancy was observed in 31 (54%) women, whereas late pregnancy infection was observed in 40 (70%) women (Table 1). Early and late parasitemia were not associated with each other (P = .89), but both were associated with placental malaria (early, P = .03; late, P = .01). The lack of correlation between early parasitemia and late parasitemia may reflect the fact that chemoprophylaxis began at enrollment and the earliest IRS exposure was at 20 weeks gestation.

Table 1.

Demographic Data^a

	All Pairs	Cord Blood MMc Negative	Cord Blood MMc Positive
Characteristic	N = 57	N = 38	N = 19
Maternal age	21.3 (16.1–31.2)	21.4 (16.2–31.2)	21.0 (16.1–29.2)
Gravidity
Primigravidae	38 (67%)	25 (66%)	13 (68%)
Multigravidae	19 (33%)	13 (34%)	6 (32%)
Delivery estimated gestational age	39 (35–42)	39 (35–42)	39 (36–41)
Male sex	31 (54%)	21 (55%)	10 (53%)
Malaria infection
No malaria	8 (14%)	6 (16%)	2 (11%)
Early only	9 (16%)	6 (16%)	3 (16%)
Late only	18 (32%)	14 (37%)	4 (21%)
Early and late	22 (39%)	12 (32%)	10 (53%)
Symptomatic malaria	18 (32%)	13 (34%)	5 (26%)
Placental malaria	31 (54%)	21 (55%)	10 (53%)

Abbreviations: MMc, maternal microchimerism.

^aNumbers presented are median (range) or n (%); percents may not add to 100% due to rounding.

Predictors of Cord Blood Maternal Microchimerism

We first assessed the distribution of total β-globin gEq assessed as well as the levels of microchimeric gEq detected in cord blood. The mean number of gEq assessed in cord blood was 39 409 (standard deviation [SD] = 16 858). One sample had a calculated level of 22.9% maternal chimerism confirmed by qPCR targeting 2 different maternal-specific alleles. This mother was LAMP positive at 28, 32, 36, and 40 weeks as well as placental malaria positive. Because this cord blood was also uniquely LAMP positive, we concluded that this sample represented a large transplacental transfusion event rather than physiologic MMc and excluded it from further analysis. Of the remaining 56 samples, 18 (32%) were positive and the mean level of all samples was 6/100 000.

Plasmodium falciparum infection during early pregnancy was associated with increased detection of cord blood MMc in adjusted analyses (adjusted odds ratio [AOR] = 3.91; 95% confidence interval [CI], 1.15–13.29]; P = .03), whereas late infection was not (AOR = 1.27; 95% CI, 0.33–4.92; P = .73) (Figure 1, Table 2, Supplementary Figure 1). Likewise, there was a trend toward increased level of cord blood MMc in offspring of women with early parasitemia (DRR = 5.52; 95% CI, 0.84–36.25; P = .08) (Figure 1, Table 2), but this was heavily influenced by a single cord blood with very high MMc (261/100 000). No relationship between early parasitemia and level of cord blood MMc was observed when this sample was excluded (Table 2). Neither symptomatic malaria during pregnancy nor placental malaria was associated with detection of MMc (Table 2).

Figure 1.

Maternal microchimerism in the cord blood stratified by early peripheral parasitemia in the mother. Maternal parasitemia detected by loop-mediated isothermal amplification (LAMP) at enrollment (12–20 weeks gestation). Maternal microchimerism (MMc) measured in 56 infants. Early maternal parasitemia was associated with increased detection of MMc (adjusted odds ratio = 3.91; 95% confidence interval, 1.15– 13.29; P = .03). gEq, genomic equivalents; ND, not detected.

Table 2.

Predictors of Cord Blood Maternal Microchimerism

	Odds Ratio for Detection		Detection Rate Ratio
Covariate	[95% CI]		[95% CI]
	Unadjusted	Adjusted	Unadjusted	Unadjusted High Sample Excludeda	Adjusted High Sample Excludeda
Early parasitemia	OR = 3.04 [0.86–10.77], P = .09	AOR = 3.91 [1.15–13.29], P = .03	DRR = 5.52 [0.84–36.25], P = .08	DRR = 0.84 [0.29–2.42], P = .75	DRR = 1.30 [0.52–3.28], P = .6
Late parasitemia	OR = 1.21 [0.35–4.19], P = .8	AOR = 1.27 [0.33–4.92], P = .73	DRR = 3.74 [0.55–25.23], P = .2	DRR = 0.73 [0.21–2.52], P = .61	DRR = 1.05 [0.76–1.44], P = .8
Symptomatic malaria	OR = 0.71 [0.21–2.42], P = .6		DRR = 0.19 [0.03–1.19], P = .08	DRR = 0.85 [0.26–2.76], P = .8
Placental malaria	OR = 0.81 [0.26–2.50], P = .7		DRR = 0.12 [0.02–0.74], P = .02	DRR = 0.74 [0.26–2.10], P = .6
Primigravidae	OR = 0.97 [0.29–3.20], P = 1	AOR = 0.80 [0.24–2.65], P = .7	DRR = 0.09 [0.02–0.48], P = .005	DRR = 0.40 [0.16–1.02], P = .05	DRR = 0.42 [0.15–1.16], P = .1b,c
Delivery estimated gestational age	OR = 0.85 [0.57–1.26], P = .4		DRR = 0.76 [0.25–2.31], P = .6	DRR = 0.94 [0.66–1.34], P = .7
Male sex	OR = 1.02 [0.32–3.24], P = 1		DRR = 6.53 [1.05–40.51], P = .04	DRR = 1.06 [0.37–3.06], P = .9
Indoor residual spraying	OR = 1.91 [0.55–6.66], P = .3	AOR = 2.55 [0.74–8.81], P = .1b,c	DRR = 9.87 [1.61–60.62], P = .01	DRR = 2.02 [0.65–6.29], P = .2	DRR = 1.81 [0.68–4.78], P = .2c

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; DRR, detection rate ratio; OR, odds ratio.

^aSingle high sample with level of 261/100 000.

^bCovariate included in adjusted model due to prediction.

^cCovariate included in adjusted model due to confounding.

Predictors of Infant Maternal Microchimerism

Next, we sought to describe the effect of age and first parasitemia on MMc levels during infancy. Twenty-nine of 57 infants experienced at least 1 episode of parasitemia during clinical follow-up, all of whom had fever. For 22 of these infants, paired samples were available from the time of first parasitemia and from a preceding timepoint. First parasitemia samples were obtained at a median age of 550 days (range, 364–962 days), and the median interval between the 2 samples was 20 days (range, 3–61 days before parasitemia sample). The mean number of gEq assessed was similar between the preparasitemia sample (42 809; SD = 13 149) and the parasitemia sample (41 444; SD = 10 869). Of the 22 infants, 6 (27%) were positive for MMc in the cord blood with a mean level of 13.2/100 000, 5 of 22 (23%) were positive for MMc in the preparasitemia sample with a mean level of 6/100 000, and 3 (13%) infants were positive for MMc during first parasitemia with a mean level of 1/100 000 (Figure 2). Overall, 12 (55%) of 22 infants were ever positive for MMc.

Figure 2.

Maternal microchimerism (MMc) across infancy. Maternal microchimerism from cord, preparasitemia, and first parasitemia samples for n = 22 infants observed during the first 3 years of life. Maternal microchimerism detected in 55% of infants overall. Maternal microchimerism positivity at any timepoint was not associated with positivity at any other timepoint. Maternal microchimerism did not significantly decay over time with the exception of 1 individual whose level declined from 261 to 101 to 0/100 000 across the 3 timepoints (lavender line). Maternal microchimerism was not different between preparasitemia and parasitemia samples when this individual was excluded (IRR = 0.4; 95% confidence interval, 0.1–2.0; P = .3). gEq, genomic equivalents; ND, not detected.

Among these 22 infants, we first asked whether the prevalence or level of MMc varied between the cord blood and preparasitemia samples to understand whether there was decay over the first 2 to 3 years of life. Within individual babies, there was no association between cord blood MMc positivity and MMc positivity at the preparasitemia timepoint (P = .7). At a group level, there was no significant difference in the odds of detection of MMc between the cord blood and preparasitemia samples (OR = 0.78; 95% CI, 0.18–3.41; P = .8). The level of MMc was lower in the preparasitemia samples relative to the cord blood samples (DRR = 0.5; 95% CI, 0.3–0.8; P = .009), but this was heavily influenced by the single individual who dropped from 261/100 000 to 101/100 000 (Figure 2). When this individual was excluded, there was no significant difference between the 2 timepoints (DRR = 1.3; 95% CI, 0.2–6.8; P = .8). Overall, these data suggest that the detection and level of MMc was relatively stable between the cord and preparasitemia samples, with the exception of a sharp decline in level within 1 individual.

We next asked whether detection or level of MMc varied between the preparasitemia sample and the first parasitemia sample to understand whether MMc cells expanded in response to P falciparum infection. There was no association between detectable MMc at the preparasitemia sample and the first parasitemia sample (P = .6). Counter to our hypothesis, there was a nonsignificant decrease in the odds of detection of MMc at the parasitemia visit (OR = 0.52; 95% CI, 0.11–2.52; P = .4). The level of MMc was significantly lower in the parasitemia samples relative to the preparasitemia samples (DRR = 0.1; 95% CI, 0.0–0.7; P = .02), which was driven by the individual who dropped from 101/100 000 to 0/100 000 (Figure 2); when this individual was excluded no effect was observed (DRR = 0.4; 95% CI, 0.1–2.3; P = .3). These data, taken together, suggest that there was no significant relationship between the detection or level of MMc by parasitemia status, with the exception of a continued decline seen in 1 individual.

We next constructed an adjusted longitudinal model that considered the effect of maternal early and late parasitemia, IRS, and infant parasitemia to predict MMc across infancy using data from the preparasitemia and first parasitemia samples (n = 44). In the adjusted model, none of the covariates significantly predicted detection or level of MMc during infancy (Table 3). Gravidity could not be included in the adjusted analyses because no infant born to a primigravid mother had detectable MMc at either of these timepoints; in unadjusted χ ² analysis, the proportion of samples positive for MMc was significantly different by gravidity (among offspring of primigravida: 0 of 16 samples positive; among offspring of multigravida: 8 of 28 samples positive; P = .02).

Table 3.

Predictors of Infancy Maternal Microchimerism

	Odds Ratio for Detection		Detection Rate Ratio
Covariate	[95% CI]		[95% CI]
	Unadjusted	Adjusted	Unadjusted	Unadjusted High Sample Excludeda	Adjusted High Sample Excludeda
Early maternal parasitemia	OR = 1.07 [0.24–4.79], P = .9	AOR = 1.68 [0.33–8.44], P = .5	DRR = 10.1 [1.3–79.0], p = .03	DRR = 2.9 [0.5–18.6], P = .3	DRR = 1.5 [0.2–12.4], P = .7
Late maternal parasitemia	OR = 1.51 [0.16–14.18], P = .7	AOR = 1.68 [0.20–14.16], P = .6	DRR = 5.3 [0.6–46.9], P = .13	DRR = 1.5 [0.2–11.4], P = .7	DRR = 1.3 [0.7–2.4], P = .5
Primigravidae	Not calculable; χ 2: P = .02		Convergence not achieved	Convergence not achieved
Indoor residual spraying	OR = 1.58 [0.27–9.23], P = .6	AOR = 2.10 [0.30–14.74], P = .5	DRR = 2.0 [0.2–16.9], P = .5	DRR = 0.3 [0.1–1.5], P = .1	DRR = 0.5 [0.1–2.7], P = .4
Infant parasitemia	OR = 0.52 [0.11–2.52], P = .4	AOR = 0.50 [0.10–2.53], P = .4	DRR = 0.1 [0.0–0.7], P = .02	DRR = 0.4 [0.1–2.3], P = .3	DRR = 0.4 [0.1–2.0], P = .3

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; DRR, detection rate ratio; OR, odds ratio.

aSingle high sample with level of 100/100 000 at preparasitemia.

DISCUSSION

The determinants of MMc in the offspring have been little studied. In this study, we report that maternal P falciparum parasitemia in the first half of pregnancy was associated with an increased prevalence of MMc in the cord blood. We did not find any association between maternal symptomatic malaria or placental malaria and MMc. In addition, we found no association between acute P falciparum infection during childhood and an increase in detection or level of MMc.

We have previously observed an increased detection and level of MMc in the offspring of primi- and secundigravid women with placental malaria in a cohort from Tanzania, where information on peripheral infection was not available and prophylaxis with sulfadoxine-pyrimethamine was ineffective [19]. In this study, we found no association between placental malaria and MMc, suggesting that placental malaria in the 2 study sites may have been qualitatively different. In Uganda, placental malaria was diagnosed by histopathology with evidence of either parasites or pigment, with the majority of cases demonstrating pigment but no parasites, consistent with past rather than active infection [14]. In contrast, in Tanzania, placental malaria was diagnosed by the presence of parasites on placental blood smear, so all infections were active and the strongest association was with inflammatory placental malaria, suggesting that disease severity may play a role in the association with MMc [9].

Our finding of an association between early peripheral parasitemia and increased detection of MMc is consistent with the observation of maternal cells in the fetus by the beginning of the second trimester [2], suggesting that trafficking may begin as early as the first trimester. Recent work has demonstrated that maternal blood may enter the intervillous spaces as early as 6 weeks gestation [20, 21], with extensive remodeling of the spiral arteries occurring between 6 and 20 weeks gestation [22]. In addition, there is increasing evidence that early P falciparum infection has long-lasting consequences for placental development and function. Placental binding (VAR2CSA expressing) parasites have been identified in first trimester [23], and P falciparum infection early in pregnancy was associated with abnormal utero-placental-fetal blood flow [24–26] as well as poor pregnancy outcomes [23, 27–31]. Plasmodium falciparum infection before 15 weeks gestation has also been associated with a decreased volume of placental transport villi, an increased diffusion distance in diffusion vessels, and a compensatory increase in diffusion vessel surface area [10]. These alterations in placental structure and function may have a significant impact on maternal cellular trafficking across the placenta, particularly during early gestation.

Prior work in cord blood mononuclear cells from a nonmalaria-endemic setting demonstrated the highest levels of MMc in antigen-experienced T cells [5]. Further, in our prior work we found that infants with cord blood MMc were more susceptible to malaria infection but less susceptible to disease when infected [9]. We hypothesized that this might be the result of the acquisition of malaria-specific T cells, which would expand upon primary parasitemia in the infant. To indirectly test this hypothesis, we evaluated the level of MMc during first parasitemia relative to a timepoint just before first parasitemia. In contrast to our hypothesis, we did not find any evidence of expansion. This may be because (1) the fetus has acquired cells that are not malaria-specific, (2) we sampled the infants too early during infection to capture maternal cell expansion, or (3) primary infection occurred late in infancy and malaria-specific maternal cells had decayed by that point.

We subsequently used data from the preparasitemia and parasitemia sample to model predictors of MMc during infancy in an adjusted model. The only notable finding was the absence of MMc in any samples from infants of primigravid mothers, a finding not previously described. First time pregnancy is associated with a higher rate of pregnancy loss [32] and pre-eclampsia [33], which may reflect lower maternal-fetal tolerance, where such tolerance may be necessary for persistent MMc at delivery. This will need to be validated in future larger studies.

Overall, more than half of infants had MMc detected either in cord blood or at one of the infancy timepoints, and there was no association between timepoints. For example, some infants had undetectable cord blood MMc but were positive at the preparasitemia or first parasitemia timepoint or both. This observation is consistent with levels of MMc that may fluctuate in the peripheral blood just above and below our limit of detection. We have recently found in a South African cohort that MMc levels increased across the first 6 months of life and were associated with breastfeeding (Balle, unpublished), suggesting that infants may acquire maternal cells via the breastmilk in addition to in utero and that breastmilk derived cells may have unique antigen specificity [34, 35].

Our study has several limitations. First, as a nested cohort study, we were restricted to the maternal-infant pairs with samples available. Our rate of identification of an informative marker of MMc was similar to other cohorts; however, we were not able to assess MMc in 30% of the subjects. Infants were enrolled in a chemoprevention trial, and an IRS campaign occurred after infant enrollment [36], resulting in an unexpectedly low rate of infant parasitemia, with only approximately half of infants infected during the follow-up period. Nonetheless, the study design in which infants served as their own control allowed us to compare MMc levels just before and concurrent with first parasitemia. In addition, the majority of these infections occurred between 2 and 3 years of life, which may have impacted our ability to detect MMc. Future work should consider the level of MMc at first parasitemia in a higher transmission setting where infection is more likely to occur in the first year of life at a time when MMc levels may be higher. In addition, to the best of our knowledge, this is the first report of utilizing dried blood spots as a substrate for MMc assessment. Although dried blood spot storage may result in DNA degradation over time, using our optimized protocol [15] we were able to obtain a range of total gEq similar to our prior study from Tanzania [9]. The use of dried blood spots for such studies will significantly increase our ability to further investigate MMc from populations outside of the United States.

CONCLUSIONS

In conclusion, we identify an association between early peripheral P falciparum infection in pregnant women and increased detection of MMc in the offspring. This supports the notion that P falciparum infection during the first half of pregnancy may have long-lasting implications for fetal and infant immune development, including to the inherited maternal cellular graft. Future work should be directed at improving our understanding of the functional capacity and impact of these maternal cells on both malaria and nonmalaria outcomes.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This work was funded by the National Institute of Child Health and Human Development (5P01HD-59454-07), the National Institute of Allergy and Infectious Diseases (K08 AI135072), Burroughs Wellcome Fund (CAMS 1017213), and undergraduate research awards from the University of Washington Department of Microbiology

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.